So you do have the necessary chipset drivers installed?
"The only things installed are official drivers for my hardware, and nusb36."
What graphics drivers are you using?
When did the problems start to appear? Right out of the box or inbetween the installation of the aforementioned drivers? As there apparently have been some installations done.

I'm using Nvidia drivers 77.72.

Yeah, I've tried a few installations. The only other hiccup I've found is that chipset drivers will often cause a bsod on first reboot after installation. Booting into safe mode then back into normal seems to solve it. But maybe there is something else going on here. It's that same with Intel Chipset 6.3.1.1004 from Vogons and 6.2.1.1001 from the MSI website.

Check SMART of your HDD, if you have high and growing UDMA/SATA/CRC/ECC errors you probably have problems with either storage drivers or DMA transfers in general between your motherboard and HDD. With low numbers, lets 10-200 it usually means bad cable used someday somehow, if you will see thousands - that is DMA, overclocking of bus, unstable noisy power. And that cause lockups and freezes up to several minutes, unprecedentedly long boot times, etc.

I've enabled SMART in BIOS and it passes the simple tests on boot. Do you have a suggestion for software to look at the SMART status in more detail?

Ok, I ended up using Smart UDM from Hiren's Boot CD. The results aren't good. But they aren't the worst either. So, chipset drivers? I've tried a couple already, guess I'll go look for some more.

As part of my troubleshooting I tried installing Windows XP and received plenty of file copy errors. During XP install it would be using it's own drivers wouldn't it? Does this point to PSU?

your ram modules are bad. Run memtest86

I gave it about an hour and found no errors. I need to go longer?

Disable DMA in Windows and BIOS, switch to PIO4. Machine will boot and load software slower than expected from normal operation. If you not receive lockups proceed to drivers. If any lockups occurs then you probably having dying southbridge, you may check it with your finger must can hold it forever for a healthy one.

Oh, i missed the XP portion. Try to install it with 1 stick of ram or 2 in single channel. Sometimes memtest can't detect instability in dual channel operation. If you have spare RAM modules try with different ones, maybe current ones not working correctly with your mobo.

Random hard lockups in Win9x make me think of a resource conflict (IRQs etc). But since you don't have ISA maybe that's not an issue. I've never used Win9x on hardware as new as you're using.

Try Prime95 "torture test" for maybe 1-2 hours. I think it turns up stability issues in the CPU/RAM region more readily than Memtest does.
Memtest86 should probably run longer than 1 hour. You want it to get fully heated and go through a few full passes. I'd just leave it overnight. If you don't trust the system's cooling though then don't do this (and fix the cooling until you do).
Preferably I'd overclock the FSB slightly during stress tests. That way if it passes you can be sure it's stable at standard clocks and not just dancing on a knife edge. If it won't pass tests with a 3% overclock then it's not acceptable IMO.

Do you have a multimeter? I don't know if it's the problem here but I believe in checking voltages on any PC build.
You can check all PSU supply voltages at either the unused drive connectors, backprobing the ATX connector, and backprobing the 4pin CPU power connector. Negative probe can go to the PSU casing.
Measuring onboard voltages is more difficult.
The voltage monitor in the BIOS isn't reliable, they're all over the place versus a real meter.
Measuring voltages under load is better, especially if load seems to have anything to do with issues that are occurring.
People are often tempted to parts swap instead of test, but if you are able to measure the voltages you'll have better knowledge than what you get by swapping parts.

What happens if you pull one stick of ram.

People are often tempted to parts swap instead of test

There are so much hidden and not obvious incompatibilities due to manufacturers faults. Swapping is the most time saving and probably effective method. And doing OC or torture test definitely not required on a hardware that refuse to perform basic tasks in 98 or during XP installation but working fine in DOS and memtest. It is not the same with system working fine but sometimes once in a week lock ups or BSOD and if you trying to reproduce what happens it doesn''t until another week passed.

"Once upon a time" i had a pretty nice system: Intel D875PBZ, Pentium 4 Northwood 3.0 and Kingston HyperX DDR-3500 2 kits of 2x1 GB. All parts bought in different places. One day motherboard died. CPU refuses to go past BIOS post in almost everything available except salvaged OEM ECS mobo from trashed PC, RAM only worked in single channel mode in every motherboard i had access to. And that is just one story...

Had similar issue -- freeze after attempting to play startup sound.
Turned out to be bad sound card drivers (by Diamond). Installing reference ones instead (from Aureal) resolved the issue.

Maybe you can clean install Win98 on system without extra devices plugged in, and gradually add new pieces of hardware (GeForce, Audigy...) and drivers -- to track down the culprit.
I know it can be PITA, but...

My guess will be the GPU -- such new cards may not be completely compatible. But it's just a guess.

So far I have ran Prime95 for two hours with no errors found.

I removed the soundcard, replaced the gpu, replaced the ram and used different slots.

Chipset drivers install without a blue screen for the first time but shortly afterwards I started to face the same kind of freezing and eventually can't get past the windows 98 boot screen.

Motherboard? Any way to confirm it? Is it already confirmed?

First of all:
1) PSU inspection: open, clean, visual check, stability test - psu tester (make sure this one works stable and properly/ heart of every computer, not the CPU, CPU is a brain)
2) Hardware disassembly, visual inspection, maintenance, clean expansion cards connectors/ram modules with ereaser - to get it nice and shiny for proper connectivity, change all new thermal compound.
If all looks good then Put it back together.
3) use post card if you have one, alternatively use software to check system stability, use software to check hdd condition
it's a lot of fun to troubleshoot. you can test cpu/gpu individually to see if them works properly.
Remember overclocking/ESD might cause unstable issues too (with used components you don't know how previous owner treated his hardware)
4)if all passed sucessfuly check drivers issues and functionality

Hopes that's clarify a bit,
Best Regards,

Godlike

Chipset drivers install without a blue screen for the first time

miss BSOD in previous answers, my bad, is it happens if you install chipset drivers and reboot after it before proceeding with further ones, or you bulk install everything? (i had similar issue with Creative sound cards, it was demo-registration program it loads first time after installing software package and reboot, if you remove it in registry before rebooting there are will be no BSOD)

Last try for drivers - boot with logging and examine bootlog.txt. Also load in command prompt mode and run Scandisk after unsuccessful boot and examine what it found, expect lost files and fragments, not just incorrect free space. Try to boot in safe mode after you can't normally.

If we assume mobo is dying there are several suspects: south bridge - temperature too high to sustain putting finger on it longer than few seconds (ICH5 is notorious for being damaged due to cheap and poor implementation of protection on inputs by mobo makers), overheating north bridge (but i think with such problem it will never pass memtest), CPU socket with broken and/or cold joints (could be checked by applying external pressure to mobo back during operation). Last one could be CPU VRM, but with hours of memtest and prime it must be dead and it's hardly possible that it will contribute to continuously growing cumulative errors leading to freeze and inability to boot.

On Hiren's boot cd you must have Victoria or MHDD, low level HDD surface test. Run it, if you encounter uncorrectable errors during read, run it again with zeroing surface and repeat read test.

Did you tried single channel mode for RAM?
Did you tried to install WinXP again and got same copy errors after changing RAM and especially single channel? You may have damaged CD or dying drive who knows, but it will be informative to do this.

Replace PSU. Or measure old, but if voltages within 10% margin it wouldn't help unless you inspect it with oscilloscope, because first victim of ripple noise is always hdd.

Maybe you have spare IDE drive, try with it. If it happen to work, then maybe you have incompatibility between your current SATA hdd firmware and first gen sata implementation in ICH5. Seagate Barracuda 7200.7/8 known to have issues with nForce and Silicon Image controllers, some WD drives have issues with first gen VIA or first gen in general.

Next stop googling every option in BIOS, disabling unneeded, trying to understand what happens with IRQ (during post and after loading Windows) and address mappings, changing ACPI modes (if you have such option).

Now you may understand how it's great to work in tech support and why swapping is sometimes lifesaver. 😀

So far I have ran Prime95 for two hours with no errors found.

Chipset drivers install without a blue screen for the first time but shortly afterwards I started to face the same kind of freezing and eventually can't get past the windows 98 boot screen.

At this point software issues look likely, but I'll still talk about hardware because I believe in being thorough on that front (on any build, even if it appears to be working) before going any further with software.

Have you tried resetting the BIOS and loading defaults? I know it's a pain to do that and lose all the settings, but there are hidden settings that don't appear in the menu and if they get corrupt they can cause unpredictable problems. The only way to be sure the CMOS settings are valid is by loading defaults.
What is the voltage of the battery while system is off and when running?
I have an MSI 875P Neo-something board which I'm guessing might be similar to yours. Back when I tried using it I ran into stability problems. I remember there was some sort of "performance optimized" mode or whatever it was called in the BIOS. It made my board unstable. Do you have a setting like that? If you do, make sure you're not using it at least until the problem is resolved. In my case, my board still wasn't satisfactory even with that setting turned off, but it made a big difference.

If you have a multimeter, try measuring as many voltages as you can. See if anything is low, and if anything is sagging under load. Installing an OS is a sustained stress on the whole system that could be exposing power issues. Note that power issues aren't only caused by the PSU, they can also be caused by weak components on the mainboard which leads to the onboard regulated voltages (Vcore, Vmemory, etc) not being properly maintained. Swapping the PSU doesn't give as much confidence to draw a conclusion, but it's better than nothing. 😀

Some POST cards and a few multimeters can measure the PCI clock frequency, but that's not common. It would be nice to know for sure that it's at 33MHz.
Does this board have FSB jumpers? If so, those typically control the PCI divider so it's important that they be set accordingly. Many boards have both jumpers and a BIOS setting, but if you just set the FSB in the BIOS and don't set the jumper, you'll end up with the wrong PCI clock.

Copy errors during install are often the result of CPU/RAM instability, but then you'd probably see failures in Prime95 if that was the issue.

HDAT2 is a good utility for testing hard drives.

I'd try booting a linux CD and see how it behaves. If the boot messages or logs show any weird warning messages, search them.

A desperation experiment would be to try installing with very relaxed settings - underclock the FSB or install a slow CPU with a slow FSB.

Hello.
I Had that exact same problem with a MSI 645E Max-U. Check your bios, look for ram timings, check in windows with CPU-Z and be sure your mobo is setting the timings right according to SPD. Check for something like agp timing settings or something related to agp/ram timings and set it to normal or safe.
My 645E was automatically setting these to fastest and at the same time using CAS 2 when SPD called for 2.5. Once i manually set to Normal AGP timing and CAS 2.5, Windows 98 became 100% stable.

So what has been happening so far...

Complete:
PSU visual inspection and cleaning
Southbridge temperature check - OK
Northbrige temperature check - Seems OK by hand. I've got this actively cooled monstrosity that I hate but I should at least expect it to be cool, right?
Reset the BIOS to defaults
Redo CPU thermal paste
HDD test with Victoria - fails to initiate drive. Says "init drive... Error!". Also has error when trying to get smart attributes.
Tried an alternative HDD. 80gb, PATA this time. Fdisk says that it's only 10773Mb

Notes:
On chipset drivers - I've been just using the installer which means that bulk hardware is being installed on reboot

There is still a "PCI Universal Serial Bus" which has a ? in device manager. When installing the .inf an error saying "A device attached the the system is not functioning" appears. Installing manually causes a lock up, removing this device causes lock up, trying to update the drivers automatically causes lock up.

HDD is a Seagate 7200.9 drive (referring to _uv_'s comment on firmware compatibility)

CPU came out of the socket along with the heat sink while the lever was still down. I'm concerned

Questions:
How do I measure battery voltage while system is running?

To do:
PSU test
HDD zeroing
Run scandisk after unsuccessful boot and examine what it found
Boot with logging
Boot a Linux CD
Try IDE drive

HDD test with Victoria - fails to initiate drive. Says "init drive... Error!". Also has error when trying to get smart attributes.

Either old Victoria and can't detect further than standard IDE, or you may try "harder" and read some manuals how to negotiate with additional ports, or you switch SATA in BIOS to compatible/legacy/combined or how they call it.

On chipset drivers - I've been just using the installer which means that bulk hardware is being installed on reboot

I asked for something else 😀 Did you installed chipset drivers, reboot and install everything else OR install all you may think of chipset+graphics+sound+etc and reboot after.

HDD is a Seagate 7200.9 drive - should be fine.

CPU came out of the socket along with the heat sink while the lever was still down. I'm concerned

usual thing, before removing heat sink next time twist it a bit few times

How do I measure battery voltage while system is running?

I doesn't matter. Some motherboards can't post with dry battery, if that isn't issue, it isn't issue either if PC connected to a power source, because on proper motherboards it automatically switches to standby power from your PSU and forget about draining battery.

There is still a "PCI Universal Serial Bus" which has a ? in device manager. When installing the .inf an error saying "A device attached the the system is not functioning" appears. Installing manually causes a lock up, removing this device causes lock up, trying to update the drivers automatically causes lock up.

And this points me towards overlapping IRQs, ACPI tables incompatible with Windows 98, PCI steering and so on. And that means that you need memorize IRQ values during POST, compare it to the values in device manager, search for resource conflicts. After that go to BIOS and try to manually reallocate IRQs. If it doesn't work play ACPI modes, APM, PNP OS settings. But google it first and read it carefully. You can't damage anything, but with some settings combination you will need to reinstall OS from scratch before any further experiments.

You may also disable some onboard devices like LPT, COM, sound, modem, game port to free some resources and BIOS and OS will try to reassign some of them to the rest. You may actually do it first, before more radical things mention earlier.

If that doesn't help you may try to upgrade or even downgrade BIOS, maybe manufacturer fucked up something.

PS.😀
and do the things suggested by xefe first